Spatial scale modulates the inference of metacommunity assembly processes.

The abundance and distribution of species across the landscape depend on the interaction between local, spatial, and stochastic processes. However, empirical syntheses relating these processes to spatiotemporal patterns of structure in metacommunities remain elusive. One important reason for this lack of synthesis is that the relative importance of the core assembly processes (dispersal, selection, and drift) critically depends on the spatial grain and extent over which communities are studied. To illustrate this, we simulated different aspects of community assembly on heterogeneous landscapes, including the strength of response to environmental heterogeneity (inherent to niche theory) vs. dispersal and stochastic drift (inherent to neutral theory). We show that increasing spatial extent leads to increasing importance of niche selection, whereas increasing spatial grain leads to decreasing importance of niche selection. The strength of these scaling effects depended on environment configuration, dispersal capacity, and niche breadth. By mapping the variation observed from the scaling effects in simulations, we could recreate the entire range of variation observed within and among empirical studies. This means that variation in the relative importance of assembly processes among empirical studies is largely scale dependent and cannot be directly compared. The scaling coefficient of the relative contribution of assembly processes, however, can be interpreted as a scale-integrative estimate to compare assembly processes across different regions and ecosystems. This emphasizes the necessity to consider spatial scaling as an explicit component of studies intended to infer the importance of community assembly processes.